here's a more-or-less minimal example that under specific circumstances
the restrict block --although silenced-- produces an extra empty line in
the output. This can be a problem when those things accumulate in a
loop. The blank line occurs between "hello again" and "and another" with
the latest snapshot here.
set verbose off
vecm 2 1 LRY LRM --rc --silent
print "hello again"
b = 0
b = 1
end restrict --full --silent
print "and another"
I've noticed (with the latest snapshot) that it is possible to create a
function package without the "label" field in the spec file although the
"menu-attachment" field is given. This leads to the failure of the menu
attachment. The failure itself is understandable, but I'm wondering
whether this mistake (of the package writer) could be caught at the
creation stage? Part of the DTD check?
with the snapshot from two days ago I get a crash when a long-running
script is active in the background (GUI script editor), and then
fiddling around in the function package windows.
I think this is similar to a report a while back (some months?).
Actually I was kind of (pleasantly) surprised that I could click around
so easily while the script was running. The script had loaded and was
calling a contributed package, too.
Cannot be more specific right now, sorry.
here's another vague report: I'm having problems in an MPI block when I
try to use errorif() or funcerr() in function form, in the sense that
execution is aborted. If I replace that with a lengthier variant with an
if block and a funcerr command it works.
I wasn't able to produce a minimal example yet, in the simple examples
errorif() apparently works fine also within MPI.
here's a reproducible crash with snapshot from June 23rd on Windows:
- ran a sample script via the GUI script editor (from CSDpanel)
- close the plot window that is automatically opened
- run the same script again (do not close the script output window before)
- close the new plot window
- click in the script output window and/or hit the page-down key
Then gretl is terminated. I wasn't able to get debug output because the
shell output window goes down with it.
is there any other place except the (built-in) command reference where
all the options of the set command are listed? I think having a tabular
representation would be good, also in the user guide (perhaps a new
appendix?). I'd be happy to help.
Hi, this is on Windows:
- have a window with the code view of a gfn package open
- drag an .inp file from the file explorer to that window; the cursor
shows a "+" icon, as if it were allowed to drag/add that file
- nothing happens (as it should), because the code view window is not an
expected: the cursor should just remain as a simple arrow
I've noticed that the 'spearman' command does not store its test results
(which are printed out) in the $test and $pvalue accessors. One side
effect of this is that a user might grab wrong values there which
actually stem from whatever test was done directly before.
Also it might make use of the $result accessor to store the rank
correlation itself. I'm aware that you can get this value by doing
corr(ranking(x), ranking(y)), so maybe this part is not so important.
(But might be mentioned in the help.)
I've noticed this behavior, perhaps Windows-specific:
? eval @gretldir
Das Symbol 'C' ist nicht definiert # [undefined]
? eval $gretldir
I would have expected @gretldir to either work, or to return an empty
string, or to throw an error, but not this problem. Maybe some backslash